**An Interdisciplinary Approach to the Sustainable Management of Territorial Resources in Hodh el Chargui, Mauritania**

**Chiara Caselle 1,\* , Sabrina Maria Rita Bonetto <sup>1</sup> , Domenico Antonio De Luca <sup>1</sup> , Manuela Lasagna <sup>1</sup> , Luigi Perotti <sup>1</sup> , Arianna Bucci <sup>2</sup> and Stefano Bechis <sup>3</sup>**


Received: 28 May 2020; Accepted: 17 June 2020; Published: 23 June 2020

**Abstract:** The present study proposes an analytical investigation of the natural resources and social framework of the Hodh el Chargui region (Mauritania), aiming to offer a useful instrument for planning and management to the local authorities. The situation of the region was evaluated by means of a participatory survey carried out among the local inhabitants. The obtained results include a collection of data about population, territorial organization, access to basic education and health services, infrastructure, main economic activities, and natural resources (in terms of water, both surface and groundwater, duration and intensity of rainfalls, soil types, and vegetal resources). The survey outcomes were completed with an integrated approach based on Earth Observation (EO) data supports, such as digital elevation models (DEMs) and Landsat8 imagery. The interdependence among the different data was evaluated and discussed, with regard to the influence of the availability of natural resources on the development of agricultural activities and on the general social welfare. The results are organized in the form of digital maps and a user-friendly webmap platform to facilitate access for all the technical and nontechnical actors involved in the project.

**Keywords:** natural resources; Mauritania; resource management; regional planning; participatory approach; EO data

#### **1. Introduction**

The regional planning and management of natural resources require us to consider the interactions among human needs, ecosystem dynamics and resource sustainability, keeping a balance between the different elements. In particular, in tropical rural or semirural areas of sub-Saharan Africa, the shortage of natural resources requires specific attention to correct planning and a proper direction of the interventions [1–4]. In these areas, the frequent drought periods, the ephemeral nature of surface water and the poorness of natural vegetation often generate severe risk scenarios. In addition, the vulnerability of rural and agro-pastoral communities living in these environments is often high due to the low adequacy of infrastructure that should provide access to the basic needs of the population (e.g., water, health care, education, transport).

The assessment of natural resource availability and sustainable use is, however, insufficient to carry out a correct decision-making and intervention process that should also include the evaluation of social, economic and cultural factors [5,6].

At a regional scale, the identification of more exposed communities, with suggestions on how to deal with risks, could help the proper direction of aid, helping the local authorities in risk-informed decision-making that favours the sustainable development of the territory. An effective planning process requires the combined consideration of environmental, technological, economic and socio-political factors.

With these purposes, the European Project "Renforcement Institutionnel en Mauritanie vers la Résilience Agricole & Pastorale" (RIMRAP) developed an interdisciplinary approach for the assessment of the vulnerability to risk of the population in Mauritania, with the aim of facilitating risk-informed intervention planning. The present study proposes a review of the results obtained in the analysis of southeastern Wilaya (Region), denominated as Hodh el Chargui (Figure 1).

**Figure 1.** Location of the Hodh el Chargui region in southeastern Mauritania.

The region, one of the poorest and most remote of Mauritania, is located about 1000 km east of the capital city, Nouakchott. As a consequence, the area suffers from weaknesses in public investments and, hence, of basic social services (e.g., schools and health centres), transport and communications. The administrative organization sees a division in 7 Moughataa (Provinces), subdivided, in turn, into 31 minor territorial units (municipalities). The area is mainly rural, being occupied by small villages (communities) inhabited by both nomadic and settled people, whose principal activities are pastoralism and, only secondarily, agriculture [7].

From a morphological point of view, the territory is characterised, in the northeastern part, by a high plateau (Dhar Plateau), with maximum heights of 420 m above sea level (a.s.l. in all the maps). The western cliff of the plateau delimitates a large flat area, with altitudes ranging between 150 and 250 m a.s.l., which corresponds to the most lively and actively populated part of the region.

The area is mainly covered by sand dunes, especially in the north of the region, where the aeolian deposits may reach thicknesses of more than 100 m. In the south of the region, the thicknesses are lower, and the deposits are mainly fixed dune fields extended in an ENE–WSW direction that progressively leave their place to the emerging rocky basement. The surface water resources of the region mainly consist of wadis, which, during the dry season, dry-up, leaving only sanded riverbeds. Very often, they are subject to an enlargement of the riverbed and to the erosion of the banks caused by occasional intensive precipitation and by degraded vegetation cover [8,9].

The climate is controlled by a seasonal alternation, with a long dry season (from October to June) and a short humid season (from July to September), with rainy precipitations that may sometimes be violent and catastrophic.

The hard climatic conditions and the poor development of agricultural activities imply a low coverage of cereal needs (on average, 30% of needs) and considerable imports of rice, oil, wheat flour and other commodities. This affects the country's balance of payments and exposes it to risks of external crisis. In addition, the rural exodus deprives the countryside of manpower and prevents the execution of the traditional mechanisms of development, maintenance and conservation of these fragile ecosystems, threatening the livelihoods of rural populations in these areas, undermining their resilience and exacerbating conflicts for natural resources [7,10].

The present study aims to approach this reality and create a reliable and accurate cartographical representation of the demographical, administrative, socio-economic features and availability of natural resources. The objective is to create a realistic framework of the human and natural resources in the Hodh el Chargui, in order to provide a reliable reference point to the local authorities for risk-informed planning of the interventions. The approach involved onsite data collection through a participatory survey proposed to the local inhabitants. The results of the survey were integrated and completed with the help of data retrieved with Earth Observation techniques.

#### **2. Materials and Methods**

#### *2.1. Onsite Survey*

The onsite survey was conducted with a participative approach, involving the local inhabitants, in order to obtain diffuse information on the local conditions. A specific questionnaire was developed, including inquiries on the population (total number and repartition between women and men), the percentage of literate people, the type of habitations, the infrastructures for water, energy, health, education, transportation and telecommunications, and the principal economic activities, with specific attention to pastoralism and agriculture. Besides this information, a more specific set of questions was directed to the analysis of natural resources, considering the duration of the rainy season, the perception of the climatic changes in the last 20 years and the features of soil, vegetal resources and water (surface and groundwater). This general questionnaire was recalibrated for the survey of the 31 municipalities and of all agro-pastoral communities with at least 300 inhabitants. In the former case, the survey was conducted in the chief towns, collecting information about the entire area of the municipalities. In the latter case, the survey is more specifically referred to as the single community, allowing for a capillary reconstruction of the features of the region. The complete question forms may be found in Supplementary Materials S1 and S2.

In addition, a specific questionnaire was prepared for the investigation of the features of water wells in the area (Supplementary Material S3); this survey was specifically aimed at the reconstruction of groundwater characteristics and their accessibility and governance. The collected information was completed and integrated with the available literature data that included similar surveys on the wells of the region [8,11].

#### *2.2. Integration of Collected Information with EO Data*

Despite the large distribution of the communities on the territory of the Hodh el Chargui, the data collected with the onsite survey only provided specific information. To complete the framework about the distribution on natural resources in the region, data were integrated through the analysis and interpretation of USGS (United States Geological Survey) Landsat imagery and with digital elevation models (DEMs) of the region.

More in detail, the Landsat images were standard Landsat-8 operational land imager (OLI) data products, radiometrically and geometrically corrected and referred to as Level-1TP (L1TP). The L1TP products are digital numbers with coefficients provided to convert data to either radiance or reflectance. The images, obtained from USGS Landsat collection service, are considered suitable for time-series analysis. The georegistration is indeed consistent and within prescribed tolerances (<12 m root mean square error — RMSE). The downloaded images are radiometrically calibrated and orthorectified, using ground control points (GCPs) and digital elevation model (DEM) data to correct relief displacement. GCPs were derived from the Global Land Survey 2000 (GLS2000) dataset. The onboard sensor reflective operational land imager (OLI) bands are 30-m resolution, and DNs are stored as 16-bit signed integers that can be linearly scaled to the top of atmosphere (TOA) reflectance. The selected images were acquired on 11 September 2015 for the rainy season and 15 April 2019 for the dry season, and their positions correspond to 199-049 and 200-049 of the Landsat repository. The Landsat-8 L1TP image digital numbers for each band were converted to top of atmosphere (TOA) reflectance, using the scaling factors stored in the metadata using QGIS SCP (semiclassification plugin, free version). Finally, dark subtraction with dark object minimum atmospheric correction was applied in order to obtain final surface reflectance Landsat 8 band stacks [12].

The DEM was the USGS EROS Archive Digital Elevation Model—Shuttle Radar Topography Mission (SRTM)—1 Arc-Second Global version, obtained from USGS EarthExplorer. The Shuttle Radar Topography Mission (SRTM) was flown aboard the space shuttle Endeavour 11 on 22 February 2000. The SRTM elevation data offer worldwide coverage of void-filled data at a resolution of 1 arc-second (30 m) and provide open distribution of this high-resolution global dataset [13,14].

These data were analysed and interpreted to obtain:


The automatic or semiautomatic definition of watershed basins starting from a DEM is a common practice in the GIS applicative world. However, particular cases as large flat areas, with complex river networks, flat terrains, crossed and looped channels and a large number of polders may create complex hydrographic conditions that make it impossible to accomplish automatic river network extraction and watershed delineation by using DEMs with automatic processes [15–17]. To face this problem, a specific procedure was developed, including the first phase of manual delineation of a river network, using as input the available literature—e.g., historical sources, reports, maps—and the USGS Landsat8 images. The second phase of the procedure consisted of a semiautomatic watershed definition with the ArcGis Watershed Delineation Tool [18,19] based on the DEM and the manually delineated river network. Results were then attentively checked to verify the correspondence between the automatic result and the real morphology suggested by the Landsat images.

For the analysis of the vegetation and the creation of the use of land maps, the Semiautomatic Classification QGIS plugin [20] was employed. The method requires the definition of reference areas with known land cover features that are used by the software to train the classification algorithm. In this application, we took as reference the data about natural vegetation from the surveys in the communities, using the Spectral Angle Map algorithm, with a threshold angle of 20◦ for the classification. The final Landsat 8 image classification includes water, bare soil/sand dunes and vegetated areas. The classification was repeated for Landsat images corresponding to the end of the rainy season and the end of the dry season to observe the respective distribution of water and vegetation covers in these two climatic end-members.

#### *2.3. Publication of Results on a Webmap for Dissemination and Evaluation*

Project results of an area necessarily pass through a cartographic representation. In recent years, dynamic and web mapping have constantly grown as a direct consequence of the development of

digital technologies and the wide diffusion of the Internet. This has created new methods of map production, instead of pure desktop GIS, making them more accessible, both technologically and economically. This new cartography has become one of the best tools for disseminating information and making it accessible to all [21]. Dynamic webmaps allow for a definition of the different dimensions of the project, involving wide, intermediate and detail scales, public and private corporate players, and different kinds of actions [22]. In this kind of project, one of the main goals is to make data accessible and easy to use by the general public and for NGO (non-governmental organizations) staff. Hence, it is very important to angle towards an application usable by the general public but, at the same time, preserve scientific rigour.

For these reasons, the main results of this project were uploaded on a webmap platform, accessible at the link www.geositlab.unito.it/rimrap, to facilitate their easy access by the local authorities and technical operators. The platform was obtained using the open-source QGIS plugin "QGIS2web" to create an html page that contained the cartographical data in an interactive and user-friendly map. This final product is accessible online but was also delivered to the local actors in an offline version in order to facilitate accessibility, even in the absence of a good internet connection.

#### **3. Results**

#### *3.1. Onsite Survey*

The map in Figure 2 shows the 31 municipalities' chief towns, the administrative boundaries of the municipalities, the 265 agro-pastoral communities and the water wells surveyed in this research and in previous studies [8,11].

**Figure 2.** Administrative maps of the Hodh el Chargui, with the surveyed municipalities, communities and water wells.

#### 3.1.1. Demography and General Information

The survey in the municipalities of Hodh el Chargui recorded 469,477 people, with 47% males and 53% females. The resulting repartition of this population on the 31 municipalities of the region is shown in Figure 3a, while Figure 3b shows the human density of each of the municipalities.

**Figure 3.** (**a**) Number of people living in each of the municipalities; (**b**) population density in the 31 municipalities of the Hodh el Chargui region.

The map in Figure 4 shows the percentages of literate people (i.e., people who have received basic school education) for men and women, respectively. The overall literate population is around 25%, with values around 30% for the male population and around 20% for the female population. The highest percentages of literate people (42% and 38%) may be found in the Moughataas of Nema and Djiguenni, respectively, in the municipalities of Mabrouk and Djiguenni itself.

**Figure 4.** Percentage of males and females and literate and not-literate people in the 31 municipalities of the region.

The features of the habitations in the agro-pastoral communities were surveyed, distinguishing between "En banco" habitations, hard dwellings, semihard dwellings, stone buildings, tents and sheds. The map in Figure 5 shows, for each community, the principal type. As can be seen in the pie chart, the majority of the communities are characterized by habitations "En banco", followed by sheds and, in a lower proportion, stone buildings. A small percentage of the communities see the prevalence of hard buildings, while none of them described tents or semihard buildings as the principal type of habitation.

**Figure 5.** Principal type of habitation in all the rural communities with at least 300 inhabitants of the region.

The survey also enquired about the availability of toilets, attesting that they are available in 60% of the communities.

#### 3.1.2. Infrastructure

The availability of infrastructure in the rural communities is an important parameter, well-representing the resilience of a society, since a capillary distribution of services in the territory means easier access for all the inhabitants of the region. The surveyed infrastructures involve water management, energy, health, education, transportation and telecommunications.

In the Hodh el Chargui, the main water resource is groundwater, since surface water usually shows an ephemeral character. The most important infrastructure for water management is, therefore, the water wells for access to drinking water for both human and animal use. In the survey, three different types of water wells were considered:


The total number of these types of wells and their activity status are reported in Figure 6.

**Figure 6.** Total number of wells, cistern wells and boreholes in the region and their activity status.

In addition, the survey collected data about springs (i.e., places where water emerges naturally from the rock or soil).

As shown in the map in Figure 7a, many of the communities have more than one water well in the neighbouring area, with a mean number of 17. However, the absence of controlled and scientific-based management and organization of the number of wells often affects their activity status.

**Figure 7.** (**a**) Maps of infrastructure for the access to groundwater; (**b**) maps of infrastructure for the management of surface water.

Despite the predominant role covered by the groundwater, a specific relevance is connected to the infrastructure for the management of surface water (e.g., dams, barriers) for agriculture purposes (e.g., irrigation). The distribution and the conditions of this kind of infrastructure are reported in Figure 7b. As shown in the map, despite a large number of structures, the conditions are often bad, affecting the real state of operation.

The infrastructure for energy mainly includes photovoltaic systems. Other energy sources were surveyed (i.e., wind systems and generators), but their presence was registered only in the municipalities of Adel Bagrou and Aoueinat Ez Bel, respectively. The photovoltaic systems, on the other hand, are present in 48% of the municipalities, while only 35% have a connection to the electricity network. The photovoltaic systems are also well developed in the rural communities, especially in the region of Nema (Figure 8).

**Figure 8.** Availability of photovoltaic systems in the communities.

The infrastructure dedicated to healthcare (i.e., hospitals, dispensaries and health units) is present in 43% of the surveyed communities, following the distribution reported in Figure 9a. Among the remaining 57% (i.e., 151 communities), 96 communities declared the availability of a means of transport to the nearest hospital in case of emergencies (i.e., handcart, car, or on the back of an animal), following the chart in Figure 9b.

**Figure 9.** (**a**) Percentages of communities with dispensaries, hospitals and basic health units; (**b**) means of transport available in the communities without infrastructure for healthcare onsite.

The education infrastructure (i.e., schools) represents one of the most important elements for the improvement of the local communities. The percentage of communities provided with a school is high (i.e., 94%). However, the conditions of the school buildings are bad in 48% of the cases (Figure 10).

**Figure 10.** Conditions of the dwellings of the schools in the communities.

Regarding transport infrastructure, the presence of asphalt roads was attested in only 23 of the 265 surveyed rural communities (Figure 11). The remaining communities are usually crossed by a natural road. These kinds of roads are often interrupted by the effect of the rainfall, for mean time periods that range between 1 day and 6 months, as reported on the map in Figure 11.

**Figure 11.** Map of availability of asphalt roads in the communities. For the communities without the presence of asphalt roads, the mean number of days of road closure caused by seasonal rainfall is reported.

Eventually, a good distribution of communication infrastructure was surveyed on the territory, at least in terms of mobile phone networks, which are present in 84.5% of the communities. Internet network, radio and television are less diffused, with percentages of 17.0%, 37.4% and 18.9%, respectively.

#### 3.1.3. Principal Economic Activities

As shown in Figure 12, the first economic activity in the majority of the communities is pastoralism, followed by agriculture. Other types of activities, such as commerce and artisanship, are, however, present, but to a lower extent. The mean percentages of men and women performing the different professions are reported in Table 1.

**Figure 12.** Number of communities that consider each economic activity, respectively, the first, the second, the third, or the fourth as important for their local economy.


**Table 1.** Repartition of the principal professions between men and women.

Hence, the main economic activity of the region is pastoralism, a sector that is well established and largely produces for export to other regions of the country and abroad. The activity is, however, strongly affected by seasonality. As shown in Figure 13a, indeed, the distance of the pastures from the communities is one order of magnitude higher during the dry season. In these periods, most of the shepherds migrate to the southern pastures, and only a small part of the livestock remains in the communities for requirements of milk, cheese, and meat (Figure 13b).

**Figure 13.** (**a**) Mean distance of the communities from the pastures during the dry and rainy seasons. (**b**) Features of the herds in the communities.

Agriculture, on the other hand, is strongly affected by the dry climate and the want of surface water resources. Hence, it is less developed, not being able to supply the needs of the region entirely. Recently, however, some experiences of local agricultural production have started. They mainly involve the production of vegetable crops, especially short cycle plants, which give their product in a few weeks and are therefore less sensitive to the possibility of lack of water.

In other frameworks with similar environmental conditions (e.g., [23]), the success of agricultural activities was mainly related to the introduction of innovative technologies, for instance, the use of specific seeds for semiarid areas or the application of innovative irrigation techniques (e.g., "drip solar" irrigation systems). Similar techniques may also allow the success of agricultural production in the Hodh el Chargui region, despite the arid conditions.

#### 3.1.4. Natural Resources

The availability of natural resources (i.e., surface water, groundwater and vegetation) mainly depends on seasonal precipitations. The rainy season starts between July and August and ends between September and October, as attested by the survey in the communities (Figure 14).

**Figure 14.** Answers of the communities about the beginning and end of the rainy season.

The survey also proposes an evaluation of the vegetal resources in the area: each community is invited to identify the principal type of natural or seminatural vegetation present in the neighbouring area (Figure 15). As can be seen, the majority of the communities (51%) identify the presence of shrub savannah, with low shrubs and grazing grass. These kinds of spaces are good for used as animal pastures during the rainy season. After long periods of drought, however, this low vegetation usually disappears, and in the late dry season, these pastures are usually completely dry.

**Figure 15.** Types of seminatural vegetation present in correspondence of the communities.

A high percentage of the communities (29%) attested the presence of a grove, similar to the shrub savannah, but with the presence of a few tall trees. Additionally, in these conditions, the long dry seasons usually dry the grass used as pastures during the rains. Only 2% of the communities describe the presence of "forest", usually in correspondence to a river of a water stream that, during the rainy season, assures a large availability of water and a more abundant growth of vegetation, with tall trees.

The absence of any kind of vegetation, even during the rainy season, or the presence of sand dunes is attested in 18% of communities.

The availability of water resources mainly depends on the groundwater that should also guarantee the hydric supply during the long months of the dry season.

Nevertheless, for the majority of the considered wells, the survey identified the existence of a season when wells run dry. As can be seen in Figure 16, this period is between April and May/June, i.e., in correspondence with the end of the long dry season. This phenomenon may be due to several reasons, e.g., excessive water extraction, long drought periods or the short distance between the wells.

The discharge of the pumps (reported for 22% of the wells) is mainly lower than 1 m3/h (23%) or in the range of 1 and 10 (57%) m3/h. Values higher than 20 m3/h are mainly related to boreholes or modern wells, emphasising the importance of modern infrastructure for water supply needs.

From a geographical point of view, clear differences in water depth can be recognised between the eastern and western parts of the region. In Figure 17, the depth of the static level for the surveyed wells is reported, highlighting in the eastern plateau a majority of wells with depth values higher than 40 m (i.e., less favourable conditions for the extraction of groundwater). In the western sector, on the other hand, values are generally lower than 30 m, and depth values are widely inhomogeneous.

This situation is coherent with the conceptual hydrogeological model proposed by [8], which includes three main hydrogeological units: the Upper Neoproterozoic hydrogeological unit, within the marine sediments that make up the ancient basement; the Mesozoic hydrogeological unit, within the Jurassic dolerites and Cretaceous sandstones that characterise the Dhar of Nema; the Quaternary hydrogeological unit, mainly consisting of aeolian sandy deposits and ancient and recent alluvial deposits, emerging in the wadis.

**Figure 16.** Beginning and end of the season when wells run dry.

**Figure 17.** Map of the depth of the static level in the water wells surveyed in this study and in [11].

Hence, the higher depths registered in the eastern area should correspond to aquifers in correspondence to the Mesozoic sediments of the Dhar plateau, while the shallow features and the inhomogeneous distribution of the measurements in the western area confirm the existence of multiple aquifers within the recent aeolian and alluvial deposits, with higher geological variability.

In addition to the quantitative features of groundwater resources, the survey also provided information about the water qualitative parameters, collecting both objective (e.g., electrical conductivity) and subjective (e.g., taste, smell, colour) ones.

The electrical conductivity provides important information about the salinity and, consequently, the drinkability of groundwater. For human use, no WHO health-based guideline value was established; however, a value of 2500 μS cm−<sup>1</sup> at 20 ◦C is indicated in a WHO directive [24], reflecting what is both achievable and acceptable to consumers.

For animals, generally, higher salt content is accepted, with maximum values between 5000 and 10,000 microS/cm, depending on the type of animal [25]. Even though the majority of the surveyed wells returned values lower than 2500 microS/cm, the maximum value registered in the area is very high, up to 15,000 microS/cm (i.e., not acceptable values, not even for animals). As shown in Figure 18, these anomalously high values are randomly present in the whole region. The absence of a clear distribution is probably due to the low continuity of the water bodies and the different depths of the wells.

**Figure 18.** Map of the conductivity and the total depth of the wells surveyed in this study and in [11].

The qualitative evaluations of the water taste return values are in line with the measures of electrical conductivity, again highlighting the existence of the problem of high salinity water, as reported in Table 2.

**Table 2.** Qualitative classification of water salinity in wells, based on the subjective evaluation of taste.


#### 3.1.5. Influence of Climatic Change

The rainfall and meteorological characteristics of the two seasons of the year influence the hydrological risks. The survey on the communities investigated how these characteristics have changed in the last 10 years as a consequence of worldwide climate change. The results, summarised in Tables 3 and 4, testify to a general increase in temperatures. In addition, the rainy season was described, in the majority of the cases, as shorter and with less intense and less frequent rainfall events. As a consequence, floods (and related damages) were described as constant or in diminution, while an increase in drought events (and related damages) was registered in the majority of the communities.

**Table 3.** Answers of the communities in the survey about climate change during the last 20 years.


**Table 4.** Data about the increase or decrease of damage due to atmospheric events (wind storms, floods and droughts).


According to the results of the survey in the 31 municipalities (Figure 19), the increasing drought events mostly caused damage to vegetation and pastures. Hence, a high impact is also registered for animals in need of pastures for their feeding. Agricultural products are in the third position, presumably because of the lower diffusion of this economic activity when compared to pastoralism. Water resources (i.e., mainly groundwater) having longer charging time and, therefore, a longer response to rain were less affected. Eventually, some records of people affected by drought events were reported, even though the survey does not investigate the specific reasons.

**Figure 19.** Most affected resources by the effects of drought.

#### *3.2. Additional Analyses with EO Satellite Data*

#### 3.2.1. Watershed Basins

The watershed basins extracted with the semiautomatic procedure for the area of Nema are shown in Figure 20. The river network was manually delineated following the Landsat8 images, using different false colour composites. The ephemeral rivers are complex and ramified in the northeastern part of the map, in correspondence with the Nema plateau. In this area, the presence of altitude variations favours the water flow along definite paths and increases the energy of water, enhancing its erosive power. In the flat area on the south-west, on the other hand, the water flows slowly, creating large swampy areas and lakes that, during the rainy period, are rich in water and vegetation.

**Figure 20.** River network (in blue) and watershed basins (in red) defined for the area of Nema.

#### 3.2.2. Use of Land Cover Classification Maps

Figure 21 shows the land cover classification for the area of the Hodh el Chargui. The map distinguishes between areas with vegetation, which may be used as pastures, and unvegetated areas, consisting of bare soil or sand dunes. Tables 5 and 6 report the size of the training samples (i.e., reference data) and the accuracy of the classification for the rainy season and the dry season, respectively. The drastically different results obtained for the rainy season (Figure 21a) and the dry season (Figure 21b) confirm the strong effect of seasonality on the availability of surface water and vegetation for the pastures.

**Figure 21.** (**a**) Map of use of land classification at the end of the rainy season. (**b**) Map of use of land cover classification at the end of the dry season.


**Table 5.** Error matrix of the classification procedure for the rainy season.

**Table 6.** Error matrix of the classification procedure for the dry season.


#### *3.3. Publication of Results on a Webmap for Dissemination and Evaluation*

The online web–GIS map is accessible at the link www.geositlab.unito.it/rimrap. The data reported on the maps include:


#### **4. Conclusions**

This study proposes an analysis of the vulnerability of the local communities in the Hodh el Chargui region (Mauritania) through the collection of heterogeneous data concerning environmental and socio-economic information. Data were collected with the combined use of an onsite survey, which involved the local population through a participatory approach, and techniques of Earth Observation.

The results provide a reliable framework of the local reality in terms of demographical, administrative and socio-economic features and availability of natural resources. The analysis includes an evaluation of the repartition between women and men, the percentage of the literate population, the infrastructure for the basic needs of the population and the principal economic activities, with specific attention to pastoralism and agriculture. In addition, a more specific insight on natural resources is obtained by considering the duration of the rainy season, the climatic changes in the last 20 years, the features of the vegetation cover in the different seasons and the water resources, both surface and groundwater.

This set of data aims at the identification of the main vulnerabilities of the local population, allowing for proper territorial planning and sustainable management of the resources by means of the development of correct interventions for the improvement of the territory. These data represent, indeed, the first step to the creation of specific plans of territorial development and improvement of local communities, both at the regional and local scales. Based on the described results, a "Plan of Municipal Development" will be prepared for each of the 31 municipalities of the region, emphasizing the main vulnerabilities of the municipal territory and assessing the priority of the interventions to be realized. In the framework of the RIMRAP project, part of the funds have already been dedicated to the execution of interventions considered as urgent or priority on the basis of the collected information. These interventions mainly involve the management of water (e.g., construction or repair of dams and other surface water infrastructure, water well fixing) and food security (with the construction or improvement of cereal banks).

The general overview of the region provided by this study creates the conditions for a more thoughtful and risk-informed decision-making process, bringing an improvement of the local economy, population welfare and accessibility to natural resources and human infrastructure.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2071-1050/12/12/5114/s1. File S1: Survey on the municipalities; File S2: Survey on the communities; File S3: File on the water wells.

**Author Contributions:** Conceptualization: S.B., S.M.R.B., and D.A.D.L.; methodology: S.B., M.L., S.M.R.B., and L.P.; investigation: S.B., S.M.R.B., and M.L.; formal analysis: A.B. and C.C.; writing—original draft preparation: C.C.; writing—review and editing: S.B., M.L., S.M.R.B., A.B., and L.P.; visualization: C.C., A.B., and L.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the European Project "Renforcement Institutionnel en Mauritanie vers la Résilience Agricole et Pastorale" (RIMRAP)—Financement de la Commission Européenne: 11e Fonds européen de développement. Référence: FED/137269/ACT/MR—Contract n◦ FED/2016/373-942: "Reduction de la vulnérabilité Agropastorale et amélioration de la résilience dans le Hodh de Chargui".

**Acknowledgments:** The authors desire to thank Terre Solidali Onlus for the useful contribution in the onsite activities.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Flood Assessment for Risk-Informed Planning along the Sirba River, Niger**

**Maurizio Tiepolo 1,\* , Maurizio Rosso <sup>2</sup> , Giovanni Massazza <sup>1</sup> , Elena Belcore 1, Suradji Issa <sup>3</sup> and Sarah Braccio <sup>1</sup>**


Received: 29 June 2019; Accepted: 22 July 2019; Published: 24 July 2019

**Abstract:** South of the Sahara, flood vulnerability and risk assessments at local level rarely identify the exposed areas according to the probability of flooding or the actions in place, or localize the exposed items. They are, therefore, of little use for local development, risk prevention, and contingency planning. The aim of this article is to assess the flood risk, providing useful information for local planning and an assessment methodology useful for other case studies. As a result, the first step involves identifying the information required by the local plans most used south of the Sahara. Four rural communities in Niger, frequently flooded by the Sirba River, are then considered. The risk is the product of the probability of a flood multiplied by the potential damage. Local knowledge and knowledge derived from a hydraulic numerical model, digital terrain model, very high resolution multispectral orthoimages, and daily precipitation are used. The assessment identifies the probability of fluvial and pluvial flooding, the exposed areas, the position, quantity, type, replacement value of exposed items, and the risk level according to three flooding scenarios. Fifteen actions are suggested to reduce the risk and to turn adversity into opportunity.

**Keywords:** climate change; contingency plan; flood risk; local development plan; risk management; sustainable rural development

#### **1. Introduction**

In the first decade of this century, floods have struck 11.5 million people south of the Sahara [1]. It is not surprising that community preparedness appears in the Agenda of the African Union [2]. Nevertheless, between 2013 and 2017, official development aid spent just 0.1 million Euros on this activity against 12.4 billion Euros used for disaster risk reduction in the Subcontinent [3]. Various multilateral organizations have urged or are supporting the preparation of local disaster risk reduction and contingency plans [4,5]. At present, there is little information on the state of local planning [6]. So far, the scientific community has been engaged above all in flood vulnerability and risk assessments, important activities which are, however, difficult to coordinate with local planning. For some time, attempts have been made to improve communication between climatologists and planners, also providing to the latter simplified analysis methods [7], aiming for greater collaboration [8,9], and urging the analysts to recommend specific actions [10]. The community-based approach used in many assessments has not produced the quantitative information expected by planning [11] and seems inadequate to appreciate the risks caused by climatic changes that communities have not yet experienced [12]. Ultimately, the use of vulnerability and risk assessments in planning remains a critical point [13–16]. The systematic review of flood assessments in the sub-Saharan context produced in recent years (Table 1) reveals that just one in four identifies the exposed areas according to the

probability of flooding. The threshold (flood level or amount of precipitation) above which flooding occurs is not identified. Therefore, the assessments cannot be used for early warning. One assessment out of four identifies the exposed items, but none considers among these the infrastructures and crops. Climate and land use/land cover changes are not considered. Just one assessment out of three identifies the actions in place to reduce risk. No assessment considers the opportunities offered by floods for sustainable rural development.

**Table 1.** Sub-Saharan Africa, 2010–2019. Consistency of 23 vulnerability and risk assessments with the needs of local plans.


In summary, the problem is that the findings of the assessments are not aligned with the information requirements of planners and decision makers. One solution may be to abandon the practice of conducting assessments in an isolated manner and to develop them by coordinating the interests of several organizations. The pool assessments thus implemented would be more likely to contribute to disaster risk reduction (DRR) strategies [6]. A quicker solution would be to identify preliminarily the information required by the local plans and to establish the types of findings expected from the assessments accordingly.

The aim of this article is therefore to propose a flood risk assessment oriented at local development, risk prevention, and contingency plans. A detailed and focused flood risk assessment assists the evaluation (risk level, identification of actions) and decision-making process of risk preparation (giving priority to actions, localizing them, and planning them over time).

From here onwards, using the term 'local level', we will refer to the minimum administrative jurisdiction (the municipality) and the term 'contingency plan' will be considered synonymous with emergency plan.

The flood risk assessment is developed with four rural communities distributed along 30 km terminals of the Sirba River: Tallé (population 2603 in 2012), Garbey Kourou (4634), Larba Birno (4713), Touré (4065). All communities are on the left bank and belong to the municipality of Gothèye, Niger. The Sirba has a transboundary watershed of 39,138 km2, whose upstream part is in Burkina Faso (93%) and terminal part is in Niger (7%) (Figure 1).

**Figure 1.** The transboundary watershed of the Sirba River and the four communities.

Although these communities are primarily exposed to drought [22], they were flooded by the Sirba in August 2012, by heavy rainfall in July 2018, and by the backwater of the Niger River in January 2019 [43].

The risk assessment considers the risk (R) as 'the probability of occurrence of hazardous events or trends multiplied by the consequences if these events occur' [44]. The risk is therefore the product of the hazard (H), or 'the potential occurrence of a natural or human-induced physical event or trend, or physical impact, that may cause loss of life, injury, or other health impacts, as well as damage and loss to property, infrastructure, livelihoods, service provision, and environmental resources' and the potential damages (D): R = H × D. The damages from flooding have already been considered as a determinant of the risk in Niger [45], in the Global South [46–51], and in the OECD member states [52,53]. The equation used is an alternative to the one that includes exposure, vulnerability, and adaptation. Potential damages constitute the ultimate effect of exposure, vulnerability, and adaptation [54]. The damage calculation is done on the reconstruction or/and replacement cost of each item exposed to flood. Thus, the assessment considers only the tangible, direct costs that could be generated in buildings, infrastructures, and crops.

The risk assessment is a process of definition of the scope, criteria, comprehension of the context, identification, analysis, and assessment of the risk [55]. As a result, the assessment is organized into four phases.

The first phase identifies the information that should be used in planning according to national guidelines for the preparation of local plans and according to literature [56]. This makes it possible to establish the findings expected from the risk assessment. The flood risk levels used and the local plans required by law in Niger are therefore ascertained.

The second phase identifies the flood risks (sources, causes, and events) and the resources held by each community (capacity and assets) to address them.

The third phase is dedicated to analysis. The probabilities of fluvial and pluvial flooding are calculated and the consequent flood zones are mapped, the individual exposed items are identified, their replacement value is estimated, and the risk level is determined.

The fourth and final phase identifies the possible risk reduction actions, considering the ongoing adaptation efforts developed by each community.

Risk treatment (selection of actions, distribution over time, and implementation) is not the aim of the assessment as this belongs to the planning phase.

The findings of the assessment are obtained integrating local knowledge with methodologies and information still little used in sub-Saharan Africa but which are promising for their quality and appropriateness to local development, risk preparation, and contingency plans. In the next sections, the materials and methods, results achieved, discussion, and conclusions are presented.

#### **2. Materials and Methods**

The assessment of the fluvial and pluvial flooding risk in Tallé, Garbey Kourou, Larba Birno, and Touré uses local and technical knowledge according to four phases.

The first phase compares the results of the systematic review of the published vulnerability and risk assessments with the information required by the national guidelines for the preparation of development plans in Benin, Burkina Faso, Cameroon, Madagascar, Mauritania, Niger, Senegal, South Africa, Uganda, and that used in a selection of local plans. This first step highlights the outputs required from a planning-oriented risk assessment. As regards the risk criteria, the three flooding scenarios used by the Niger Basin Authority were considered: frequent (yellow), severe (orange), and catastrophic (red). The categories of local plans required by law in Niger were identified by questioning the Sustainable Development and Environment National Council (CNEDD, according to the French acronym) and the Directorate General for Civil Protection (DGPC, in French).

The second phase identifies, through meetings with each community, the hydro-climatic threats, the past catastrophic events, the rainfall threshold, and the flood level over which damages are produced and local resources mobilized to address them (Figure 2).

**Figure 2.** Flood risk assessment flowchart.

The third phase firstly calculates the probability of occurrence (1/return period) of a flood according to three scenarios of fluvial and pluvial flooding (frequent, severe, catastrophic). The probability of flooding of the Sirba and the backwater of the Niger Rivers is calculated based upon the discharge recorded respectively at the stations of Garbey Kourou in the period 1956–2018 and of Niamey in the period 1929–2019 with a Generalized Extreme Values approach [23].

The probability of pluvial flooding is calculated starting from the daily rainfall and related damages recorded by each community in a register provided by the National Directorate for Meteorology. The probability of occurrence of frequent, severe, and catastrophic rainfall is then calculated with respect to the daily precipitations recorded at the meteorological station of Gothèye (30 km away) between 1961 and 2015. Severe rainfall forms ephemeral water bodies in Larba Birno and Touré that may cause the collapse of buildings and the leakage of contents of the latrine pit.

The area exposed to fluvial flooding according to the three hazard scenarios was identified through the hydraulic numerical model developed on the HEC-RAS software in a 1D configuration [23]. The model calculates the water surface elevations for the different discharges and extends them on the riverbed geometry. The geometry was constructed on a Digital Terrain Model—DTM with 10 m of horizontal resolution, detailed by way of river cross-sections detected each kilometer and hydraulic structures. The roughness was based on riverbed granulometry and the downstream conditions considering the hydraulic levels of the Niger River. This model was calibrated with the hydraulic levels measured at the Bossey Bangou and Garbey Kourou hydrometric stations and in the communities during the 2018 rainy season. This tool allows, towards unsteady simulations conducted with measured hydrographs, the estimation of the propagation and submergence times during the fluvial flooding.

The area exposed to pluvial flooding was identified firstly with an inspection aimed at localizing the biggest ephemeral water bodies in late August 2018, when three-quarters of the annual precipitation had already been accumulated. The identification continued by taking red, green, blue (RGB) images with a 24.3 megapixel camera and RGB near-infrared (NIR) images with a specially created low-cost 5 MP optical sensor from an unmanned aerial vehicle (UAV) [55]. Two flights were made on each community on 14 and 15 September 2018 at 270 m above ground level with the RGB camera and at 120 m with the NIR camera. The collected data were processed with SfM software and orthophotos to 6 cm in RGBN and 4 cm in RGB were obtained for each community. From the RGB data, a DTM of 6 cm of resolution was generated [57]. The water was identified using the normalized differential water index (NDWI = (Green – NIR)/(Green + NIR)) [58] on the multispectral images obtained. Finally, the area surrounding the ephemeral water bodies was estimated by way of the DTM (Supplementary File 1).

The exposed items were identified by way of photo interpretation of the orthoimages at 4 cm resolution recorded by the UAV (Supplementary File 2). Finally, the areas of potential water stagnation (i.e., maximum expected extension of stagnation) were estimated on the basis of the DTM. Those images identify the latrines, granaries, boreholes, wells, fountains, photovoltaic miniplants, and rain-fed and irrigated crops that could not be identified on high-resolution satellite images (Figure 3).

The estimate of damages to the buildings is not based upon the stage-damage function, nor does it consider the flow velocity since the dwellings collapse once they are flooded, as the majority are built from crude earth masonry. The flooding duration and the depth of the water are considered in order to estimate the damages to the crops.

The replacement value of the buildings is estimated through a focus group in each community, considering the construction cost of a standard 24 m<sup>2</sup> crude earth house, a latrine, a shower, and a granary. The crop yields are estimated by the Departmental Directorate for Agriculture of Gothèye, which also monitors the prices of agricultural products on the markets. Yields and prices of millet and paddy originate from statistics of the Ministry of Agriculture [59].

The risk level is obtained by multiplying the hazard by the damages in the case of fluvial and pluvial flooding according to the frequent, severe, and catastrophic scenarios.

The fourth phase identifies actions to reduce the risk associated with the type of flooding which threatens each community and with the exposed items as proposed by the individual communities during a participative meeting held on 25–26 June 2019 and integrated, after discussion, with some recurring actions in the examined plans.

**Figure 3.** Touré, 2018. Images for exposed items' identification: Sentinel satellite image with 10 m resolution (top); satellite high-resolution image freely available from Google Earth showing (1) earth construction (center); very high resolution image taken from a UAV showing (1) hearth house, (2) latrines, (3) house entrance, (4) ephemeral water body, (5) millet (bottom).

#### **3. Results**

#### *3.1. Aligning Flood Risk Assessments with Local Plans Requirements*

The reduction of the hydro-climatic risk in Niger is essentially entrusted to the municipal development plan (MDP). From 2015, MDPs must identify the impact of climate change and actions to reduce it [60]. The local administrations are struggling to perform this task effectively. It must be considered that in Niger the municipalities are vast jurisdictions that include many rural settlements in strong expansion and multiplication. Gothèye, for example, as an administrative jurisdiction of 3600 km2, includes 146 rural settlements and over 93,000 inhabitants (2012). The rural municipalities and the consultants used by them to draw up development plans have access to scant geographical information and often do not have the tools, resources, and time to collect additional information and to process it. As a result, the risk reduction actions are identified based upon the needs expressed by the communities during rapid consultations. In the best cases, the planned actions concern the reduction of the runoff (half-moons, stone lines, trapezoidal bunds) and of riverbank erosion (gabions, tree planting), and the increase in tree cover, measures that have no resolving effects on the fluvial flooding risk.

Community adaptation plans as introduced firstly by the ANADIA project (Italian development aid) and then by the BRACED project (English development aid) are experimental, voluntary tools. Development plans for individual settlements, as tested in Togo, and risk reduction plans, quite common in Latin America, are uncommon in Niger. The latter are usually made up of a presentation report, a map of the flood-prone areas, a map of the exposed items, a zoning plan which subdivides the territory depending on the flood probability, and a regulation, which specifies what is permitted, prohibited, and worth doing in each zone. The risk reduction plans usually recommend the preparation of a contingency plan.

In future, the contingency plan, as tested on a municipal scale in Burundi, South Africa and on a scale of individual settlements in Mali, may be required by law in Niger to address the hydro-climatic hazards in the most densely populated flood risk areas. This plan should be coordinated with the early warning system (EWS) and constructed taking account of local capacities and assets. An emergency committee, a map of the flood areas according to flooding probability, a map of exposed items to guide information, the drills, and making accessible the refuge sites to households settled in flood-prone areas are recurrent components of this plan.

These components of the plans, as taken from the national guidelines of Niger and eight other African countries and from the literature, need specific information from vulnerability and risk assessments (Figure 4).

**Figure 4.** Use of vulnerability and risk assessments findings according to national guidelines for local development plan preparation in Niger and in eight other African countries.

In the nine considered countries, municipal development plans made 98% of local plans, and are a consolidated planning tool, now in its third generation. Contingency, risk prevention, and adaptation plans at the local level remain still occasional tools.

The assessment thus provides useful information firstly for the MDPs and, prospectively, for the risk prevention and contingency plans.

#### *3.2. Flood Hazards*

The four communities along the Sirba are exposed to fluvial flooding during the rainy season. Tallé is exposed to fluvial flooding also in the dry season, due to the backwater of the Niger River. Larba Birno and Touré are also exposed to pluvial flooding (Table 2, Figures 5–8).


**Table 2.** Four communities along the Sirba River. Hazard.

**Figure 5.** Tallé, 2018. Hazard and exposed items map: frequent (1), severe (2), catastrophic (3), flood-prone area, exposed house (4), well or borehole (5), field (6), built-up area (7), unpaved roads (8), and paved roads (9).

**Figure 6.** Garbey Kourou, 2018. Hazard and exposed items map: frequent (1), severe (2), catastrophic (3), flood-prone area, exposed house (4), field (5), built-up area (6), unpaved road (7).

**Figure 7.** Larba Birno, 2018. Hazard and exposed items map: frequent (1), severe (2), catastrophic (3), flood-prone area, exposed house (4), field (5), built-up area (6), unpaved road (7).

**Figure 8.** Touré, 2018. Hazard and exposed items map: frequent (1), severe (2), catastrophic (3), flood-prone area, exposed house (4), well or borehole (5), field (6), built-up area (7), unpaved roads (8).

The analysis of the historical series of the Sirba discharge identifies that a flood of 800 m3/s has a return period (RP) of 10 years and can be considered frequent, a flood of 1500 m3/s has an RP of 30 years and is severe, and a flood of 2400 m3/s has an RP of 100 years and is catastrophic. The analysis of local precipitations highlights that a daily precipitation with 90 mm of rain has an RP of 10 years; it is frequent and can cause some damage. A daily precipitation with 100 mm of rain has an RP of 17 years, and should be considered a severe event as it generates great damage. A day with 200 mm of rain has an RP of 50 years and generates such damage as to be considered a catastrophic event (Table 2).

The hydraulic numerical model identifies areas exposed to frequent (RP of 10 years), severe (RP of 30 years), and catastrophic (RP of 100 years) flooding in circumstances of homogeneity of the historical series of discharges (stationary approach). The RP reduces respectively to 2, 5, and 10 years [43] if the nonstationary approach is adopted, necessary to consider strong changes in hydrology which reflect both changes in the climate [61] and in the land use/land cover [62–64].

The Sirba flood reaches its peak in 3–9 days and returns to normal level in 4–11 days. The propagation time from Bossey Bangou (108 km upstream of the Niger River, where a hydrometric station has operated since 2018) to Touré is 20 h, to Larba Birno is 26 h, and to Garbey Kourou and Tallé is 28 h.

#### *3.3. Flood Damage*

The very high resolution orthoimages facilitate the identification of exposed items that are not visible on satellite images (such as those available for free on Google Earth) and the identification of up to one-third more exposed items. The dwellings appear to have been withdrawn from areas exposed to frequent and severe flooding, but the crops, particularly the commercial crops in Tallé, wells, and boreholes at Larba Birno are plentiful (Table 3).

The area exposed to catastrophic flooding contains the bulk of the exposed items in two out of four communities. This stock of dwellings has the floor at ground level, the door has no threshold, and the walls and roof are made from crude earth. These constructions collapse as soon as the water enters inside or rains too intensely. Boreholes, wells, and fountains without a foundation above water level cannot be used for weeks. The lengthy duration of the flood destroys the crops, once submerged.


**Table 3.** Four communities along the Sirba River, 2018. Houses in flood-prone areas.

The total amount of damage is the highest in Tallé in the event of backwater of the Niger River, followed by Touré in the case of fluvial flooding, and by Larba Birno in the case of pluvial flooding (Table 4). Buildings account for the bulk of the damages in all scenarios and communities (from 84% to 92%) with the sole exception of the flood due to the backwater of the Niger River (Table 5).

**Table 4.** Four communities along the Sirba River. Replacement value of the exposed items K €.


**Table 5.** Four communities along the Sirba River. Exposure of houses over all exposed items (% of total replacement value, K €).


#### *3.4. Flood Risk Level*

Larba Birno presents the highest risk due to a high value of exposed items in an area with medium probability of pluvial flooding (Table 6). Touré follows, with numerous dwellings in an area with medium probability of pluvial flooding. Then comes Tallé, with a high value of exposed items in an area with low probability of river flooding, and finally Garbey Kourou, with a low value of exposed items in an area with low probability of flooding.

**Table 6.** Four communities along the Sirba River. Risk level.


#### *3.5. Identification of Actions for Risk Reduction*

Community meetings have shown that the pluvial flood opens up opportunities for recession farming (okra) in Garbey Kourou, in the ephemeral pond upstream of the road, and the fluvial flood allows crops of cowpeas, corn, and squash in Touré, especially on the right bank of the Sirba River.

The proposed actions consider these results and the actions in progress, limited to an emergency committee, flood drills, and a hydrometric station at Garbey Kourou, a manual rain gauge and a river gauge in every community, along with the withdrawal of dwellings from areas exposed to frequent and severe flooding (Table 7), a flood and drought multihazard community adaptation plan (2014) in Garbey Kourou and Tallé.


**Table 7.** Four communities along the Sirba River. Ongoing adaptation actions.

Other aspects to consider are that the inhabitants settled in the flood-prone area do not want to leave the river because they go there every day to do their laundry and to wash themselves, and there is the desire to remain in the large initial lots of the settlement, in which the ancestors lived. Finally, even in the case of destruction of the house by the river or heavy rains, the tendency is to reconstruct a little further upstream to reuse the clay of the old bricks that elsewhere would not be available or too tiring to transport. In many cases, therefore, it is not access to land that pushes some inhabitants into areas at risk but rather reasons linked to daily life, the cost of transfer, and the link with the place.

The limit of catastrophic flooding (which may occur with a probability 10 times higher according to a nonstationary approach) should be reported and the platform of the boreholes, wells, fountains, and photovoltaic miniplants should be raised as a result. The ephemeral water bodies within the built-up areas should be treated by a storm water drainage. Rain-fed crops should be moved away from the riverbanks. Paddy fields should be protected, remaking the river levees.

When farmers use their own land for commercial gardens, having them set back from the riverbanks should be possible. Cassava should be preferred to other crops since it demands less water.

Any flood risk prevention plan should contain a map of the flood area, the exposed actions, a zoning plan according to flooding probability, and a regulation.

The plan should prohibit further construction and the replacement of dwellings made from crude earth with durable dwellings in the area exposed to catastrophic flooding, unless the floor is above the level of the water.

As regards the contingency plan, an early warning system should be prepared and implemented using local rural radio and by spreading door-to-door information on how to prepare for flooding for all households settled in the area exposed to catastrophic flooding.

A contingency plan always includes a refuge site dimensioned on the number of inhabitants present in the area at risk. However, the communities along the Sirba do not need this structure because during the season in which a flood may occur, the settlements are emptied and the inhabitants move near the fields in temporary settlements to follow the crops. Drills should be organized every year and the emergency committee reactivated or established (Figure 9).

**Figure 9.** Use of risk assessment for local planning.

#### **4. Discussion**

Exceptions aside, the flood vulnerability and risk assessments considered in the systematic review are deficient in identifying the flood-prone area according to the frequency of the flood and critical rain, the exposed items, and the ongoing actions and they are, therefore, of little use for MDPs and, prospectively, for risk prevention and contingency plans at local level [14].

The initial problem was firstly to align the expected findings of the assessment with the necessary information to reduce flood risk. As a result, the requirements of information for each category of local plans in use in sub-Saharan Africa and in Niger, in particular, were identified. The assessment for Garbey Kourou, Larba Birno, Tallé, and Touré was developed using methods and information based upon those needs: a hydraulic numerical model, the river discharge time series, the long-term daily pluviometric series, DTM, and very high resolution orthoimage, completed by inspections and meetings with the communities. The adopted methodology is totally replicable in order to support the sustainable development of each rural community both in Sahelian areas and in similar contexts.

In this way, it was ascertained that the waters of the Sirba and the Niger Rivers slowly reach the flood peak. This involves the loss of crops once they are submerged. The flood subsides slowly. This means that recession agriculture after Niger River backwater is not possible and that the flood does not bring benefits to Tallé, but only damage. On the contrary, the fluvial flood in Larba Birno and the pluvial flood in Garbey Kourou allow recession agriculture. In these cases, climate change and the consequent flooding constitute an opportunity—an issue still little investigated by the literature. The slow river regime does, however, have an advantage: a flood propagation time long enough to allow for the mayor and the communities downstream to be alerted with sufficient warning to react, even if the flood peak occurs at night.

The multihazard assessment (fluvial flooding with two seasonal peaks and pluvial flooding) highlights that within a river reach of just 30 km, each community is exposed to specific hazards, has different exposed items in the flood prone area, and sometime offers rural development opportunities, and thus requires equally specific measures, a result opposite to that found in other studies [14].

Commercial crops are the main exposed items in the zone with high probability of fluvial flooding for all considered communities and, in the case of Tallé, also for the backwater of the Niger River.

Commercial agriculture is mainly carried out by women and is of crucial importance for rural households. Revenues obtained from the sale of cassava, potatoes, cabbage, squash, and tomatoes make it possible to compensate for the losses of rain-fed cereal crops in the event of agricultural drought and to sustain health, education, and clothing costs. The importance of damage to commercial crops that we have highlighted constitutes a paradox on which the literature has not sufficiently dwelt: To guarantee sustainable rural development in a semiarid context, it is essential to consider the risk of flooding as well as the risk of drought. However, the actions to protect commercial crops from flooding are expensive: the further away you go from the flood-prone area, the more need for wells whose realization, in the specific context of the Sirba, characterized by frequent rocky outcrops, can be very expensive.

The current assessments do not quantify the damages in such a detailed way due to the low-resolution images used, which prevent the individual exposed items from being localized.

The presence of many dwellings in the area exposed to catastrophic flooding, thus of many households, is scantly evidenced in the literature [24,32].

There are few risk reduction actions in progress. As to the actions to be developed, the list is longer and more specific (information, awareness, contingency plan, drills, basement/platform rise, storm water drainage, flood signals, prohibition of house consolidation, field metal fences) than that proposed by the assessments considered which mostly recommend early warning systems [20,21,29,38], resettlement [18,27,30], dam and levee construction [29,31,38], and house improvements [21,40]. These actions are not sufficient in the case study since they do not refer to specific hazards, exposed items, and plan categories.

The assessment identified the discharges at the upstream hydrometer located at Bossey Bangou, close to the Niger–Burkina Faso border, beyond which damages may occur downstream and their probability of occurrence linked the three hazard levels (frequent, severe, catastrophic) to the early warning.

The assessment demonstrates the potential of very high resolution DTM and orthoimages still little used in vulnerability and risk assessments at the local level south of the Sahara. This technology reduces the Information Technology divide if used by local firms using low-cost sensors. In addition, it offers information on the flood-prone areas and on the exposed items which high-resolution satellite images are unable to provide or which require lengthy and costly inspections.

The method that combines local knowledge and technical knowledge presents at least three advantages: it provides a complete photograph of each settlement, which inspections are unable to produce, it is feasible even in the sweltering months (April–May) during which it is not possible to perform lengthy inspections, and it provides a quantitative basis for risk monitoring. Nevertheless, the assessment does have two limits. First of all, the replacement value of the exposed items does not consider domestic vegetable gardens, plentiful above all on the right bank at Larba Birno, the service interruption of the boreholes, the wells, the fountains, the photovoltaic miniplants, or the impacts of ephemeral water bodies on health. Second, the value of production in recession agriculture was not calculated because the assessment took place too far after the last major flood (2012) and therefore the fields cultivated in this mode were no longer observable. These aspects should be addressed by future assessments. The estimate of damages is thus conservative. In addition, in the space of a few years, more high-value exposed items could be found in the area at low risk of river flooding in the absence of actions to limit their proliferation.

The findings of the assessment are above all of interest for local development, risk prevention, and contingency planning, since they provide precise quantities in defined locations and for a localized

population in order to decide on a set of proposed actions. Secondly, the results of the assessment are used to validate the hydraulic numerical model: the detailed identification of the exposed items and the flooded areas allows for the optimization of their calibration and the observation of changes following future floods [65].

#### **5. Conclusions**

In Africa, south of the Sahara, flood risk assessments at the local level do not provide information aligned with those required by local development plans, risk prevention plans, and contingency plans. A recurrent case concerns the detail of the information provided by the assessments, which is often more appropriate to the regional scale than to the local one. Furthermore, the areas exposed to flooding, the items they contain, and the proposal of specific actions to reduce the risk of flooding appropriate to the particular characteristics of the investigated contexts are too often absent from the assessments. Assessments are therefore not very useful to support an informed decision-making process in risk reduction. To overcome this problem, the simplest solution is to firstly identify the information required by local plans and consequently establish the type of results that the assessment should achieve. Furthermore, this involves reviewing the assessment approach, which must consequently integrate local knowledge with scientific knowledge, for example, by supporting the local ITC (UAV images) which provide more detailed information than anything obtained through commercial high-resolution images.

The flood risk assessment developed along the Sirba River considers and meets the needs of local planning as it establishes the probability of occurrence according to the flooding scenarios established by the Niger River Basin Authority; it consequently identifies the flood-prone areas and exposed items, assesses the potential damages and risk level, and identifies 15 risk-reduction actions to be proposed to the decision makers. Contrary to conventional assessments, this work identifies the opportunities for rural development offered by an event generally considered adverse, such as flooding. The use of very high-resolution images, thereby limits the inspections to a few, targeted receptors, saving time.

The multihazard approach, the integration of local and scientific knowledge, and the techniques used to measure the level of risk and its determinants illustrated in this case study can inspire all those flood assessments on a community scale that must be realized in contexts where information is scarce.

**Author Contributions:** Conceptualization, M.T.; Methodology, M.T., M.R., and E.B.; Validation, G.M. and S.I.; Investigation, M.T., G.M., E.B., S.I., and S.B.; Writing—original draft preparation, M.T. and G.M.; Writing—review and editing, G.M.; Visualization, S.B.; Funding acquisition, M.T.

**Funding:** This research was funded by the Italian Agency for Development Cooperation, by Institute of Bioeconomy (IBE)-National Research Council of Italy (CNR) at Florence (leader), by the National Directorate for Meteorology of Niger (DMN), and by DIST-Politecnico and University of Turin within the project ANADIA2.

**Acknowledgments:** We would like to thank Vieri Tarchiani (IBE-CNR) and Katiellou Gaptia Lawan (DMN) for facilitating the field activities, Moussa Mouhimouni (DMN) and Mohamed Housseini Ibrahim (Directorate National of Hydraulics) for the information, Aziz Kountché (Africa Drone Service) for the UAV activity, Abdou Hidjo (chief of Garbey Kourou), Salou Mounkaila (chief of Larba Birno), Harouna Mounkeila (chief of Tallé), Issa Sirfi (chief of Touré), and Oumarou Issaka (Departmental directorate for environment, Gothèye) for participation at meeting discussion.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
